DFB1r5 Design Review Thread

General discussions or ideas about hardware.
User avatar
Badwolf
Posts: 2231
Joined: Tue Nov 19, 2019 12:09 pm

Re: DFB1r5 Design Review Thread

Post by Badwolf »

Well after a bit of a delay whilst I played with SDRAM some more on the STE, I’ve put in the rev5 order as I don’t think anything from that experiment will come across now.

So fingers crossed this is finally the no bodge wire version! Just wait the month or so for delivery now…

BW
DFB1 Open source 50MHz 030 and TT-RAM accelerator for the Falcon
DSTB1 Open source 16Mhz 68k and AltRAM accelerator for the ST
Smalliermouse ST-optimised USB mouse adapter based on SmallyMouse2
FrontBench The Frontier: Elite 2 intro as a benchmark
User avatar
Badwolf
Posts: 2231
Joined: Tue Nov 19, 2019 12:09 pm

Re: DFB1r5 Design Review Thread

Post by Badwolf »

I've run into an unexpected problem with the DFB1r5 build. It's not made to the video series yet, but I got everything working (CPU at 50MHz, TT-RAM and FPU at 40MHz), but during soak testing I'm finding the machine becomes unstable as it warms up.

I've blue-tac'd my multimeter's thermocouple to the heatsink on my ceramic CPU and it seems pretty consistent that as the heatsink temperature passes about 50/51C, things stop working.

At 50MHz I'm getting about 10 minutes of full stability from cold before starting to see problems.

At 36MHz it takes longer, but still once it gets up to that temperature: trouble.

So it looks like a heat dissipation problem, yes? But the form factor is the same as DFB1r4, and that didn't experience this.

In fact running my (plastic) CPU in the rev 4 at 36 MHz saw everything nice and stable even when the heatsink temperature plateaued at 58C.

So need to do some more tests, swapping the CPUs over and measuring in-situ temperatures in r4 as well as r5, but I thought I'd ask if there are any obvious design or assembly issues that may be causing this. Wrong decoupling capacitance? Poor socket contacts? Anything of that ilk?

If I can't find anything obvious, the next step would be to build up a second copy of the board, I suppose.

Cheers,

BW.
DFB1 Open source 50MHz 030 and TT-RAM accelerator for the Falcon
DSTB1 Open source 16Mhz 68k and AltRAM accelerator for the ST
Smalliermouse ST-optimised USB mouse adapter based on SmallyMouse2
FrontBench The Frontier: Elite 2 intro as a benchmark
User avatar
Badwolf
Posts: 2231
Joined: Tue Nov 19, 2019 12:09 pm

Re: DFB1r5 Design Review Thread

Post by Badwolf »

Badwolf wrote: Fri Mar 25, 2022 3:58 pm If I can't find anything obvious, the next step would be to build up a second copy of the board, I suppose.
A weekend of testing this and I haven't got much further. My current suspect is the CPLD. My normal supplier was out of stock when I did this order so I've used a chip of dubious provenance.

The bony finger of suspicion points to that shady eBay purchase. I managed to get a repeat failure situation and then some freezer spray seemed to resolve it. I've not been able to repeat that test freezering the CPU yet, but I'm increasingly confident that the board is not fundamentally cursed!

BW
DFB1 Open source 50MHz 030 and TT-RAM accelerator for the Falcon
DSTB1 Open source 16Mhz 68k and AltRAM accelerator for the ST
Smalliermouse ST-optimised USB mouse adapter based on SmallyMouse2
FrontBench The Frontier: Elite 2 intro as a benchmark
User avatar
Badwolf
Posts: 2231
Joined: Tue Nov 19, 2019 12:09 pm

Re: DFB1r5 Design Review Thread

Post by Badwolf »

So I was still experiencing this odd apparently heat-related issue with board one.

Fortunately, board two (built for Frank, an early backer) works perfectly and is long-term stable using the same CPU (below).

IMG_5227.jpeg
IMG_5227.jpeg (201.7 KiB) Viewed 1756 times


So, I conclude it's most likely a flaky solder joint or the CPLD. I reflowed everything I could -- no improvement. OK, CPLD looks like the culprit.

So I sweat a known good CPLD off an old, cannibalised revision and solder that on.

Same problem.

Now I'm confused. Could there be a problem with the board itself? Bit of flex after it warms up means a contact is lost? Well, let's use the board I was meant to be sending to a beta tester, and build board 3.

So I sit down for two hours of soldering. Half way through the passives, I notice there's a capacitor that's not on my list. That's odd, I think. I must have missed that when I did the print out. I'd better check the BOM later.

Anyway, carry on, solder on the CPLD. Solder on my last PGA168 socket, solder on the oscillator holder, solder on the pin sockets for the Falcon's connection.

Nearly done, just the four clock resistors on the top and then it's off to put it in the Falcon.

Wait. Where are the pads?

It's then when I see the silkscreen has 'r4' after the name.

IMG_5228.jpeg
IMG_5228.jpeg (248.33 KiB) Viewed 1756 times


FML. :lol:

How do I desolder a PGA168..? :?

BW
DFB1 Open source 50MHz 030 and TT-RAM accelerator for the Falcon
DSTB1 Open source 16Mhz 68k and AltRAM accelerator for the ST
Smalliermouse ST-optimised USB mouse adapter based on SmallyMouse2
FrontBench The Frontier: Elite 2 intro as a benchmark
User avatar
sporniket
Posts: 956
Joined: Sat Sep 26, 2020 9:12 pm
Location: France
Contact:

Re: DFB1r5 Design Review Thread

Post by sporniket »

Badwolf wrote: Sat Apr 02, 2022 7:59 pm How do I desolder a PGA168..? :?
lets_play_a_game.jpg
lets_play_a_game.jpg (24.53 KiB) Viewed 1723 times
User avatar
kodak80
Posts: 370
Joined: Sat Oct 21, 2017 1:14 am
Location: Brisbane,QLD Australia
Contact:

Re: DFB1r5 Design Review Thread

Post by kodak80 »

Badwolf wrote: Sat Apr 02, 2022 7:59 pm How do I desolder a PGA168..? :?
BW
you could cover the IC with thermal kapton (heat resistant) tape and use a heat gun to remove it. Hopefully, the tape will protect the IC.
Atari Falcon 030 | Atari 1040 STE | Atari 1040 STFM | Atari 1040 STF | Kryoflux & Supercard Pro Flux boards
Creator of the Atari ST Review magazine archive: https://www.chillichai.com/atari-st-review
Steve
Posts: 2570
Joined: Fri Sep 15, 2017 11:49 am

Re: DFB1r5 Design Review Thread

Post by Steve »

Put the socket on the wrong side? Cover the rear pins with some good rework flux (the kind that withstands high temps) put a heat gun on the area at about 370c, swirl the heat around and the socket should just 'drop' out. Then clean up with wick.

If you don't have a heat gun it's gonna be a long job...

The socket may warp by the way, so may be unusable after... the faster you get it to 'drop out' the better.
User avatar
Badwolf
Posts: 2231
Joined: Tue Nov 19, 2019 12:09 pm

Re: DFB1r5 Design Review Thread

Post by Badwolf »

Some days it doesn't rain, but it pours.

I was fortunate, I thought. I found an extra PGA128 (not sure where I got 168 from) socket in a box of chips. Must have forgotten to put it in drawer with the rest. So I'll start building up the proper board.

Things are going well, I get through the basics in double-quick time stealing the components directly off the errantly assembled r4 board.

The CPLD has come off and gone on in a surprisingly easy manner for a desoldered one.

I even had a go desoldering the pin socket strips as a bit of a test of getting off the PGA socket. It kind of worked for one, failed for the other, but hey, progress of sorts.

I do my normal thing of buzzing out the various power rails to make sure there are no obivous shorts. Wire up the bench supply with a 200mA current limit to make sure we don't see any clipping. We don't, but it's at the high end of what's acceptable.

I scratch my chin, get out the magnifier, suck my teeth a bit.

...then realise I've put the 5V jump wires on the 3V3 side. Another CPLD bites the dust! :oops:

Balls. What a mess this day has been. I can't face desoldering and resoldering another CPLD so I'll go back to board one and see if I can relfow it a third time with any success.

This time I'll try putting the heat gun on the PGA pins and making sure there's a nice flow on each pin.

I flux it up, set the gun to 360C and do my thing. None of the pins seems to be flowing much at all. The big old hunks of metal going through the board must be sinking a lot of heat. Bit of smoke produced -- must be the flux burning, I'll ease off a bit.

I didn't seem to get very far with that. Never mind, call it quits. I'll clean up and go back for a soak test again.

Turn the board over... Half the socket has melted and the pins are all covered in goo and misaligned now. That must have been the smoke... Chalk off another socket. And likely that's the end of board 1 too. Double FFS. :oops:

I decided to have Sunday off! :evil:

At least it's shown reclaiming PGA sockets isn't going to be straightforward. ;)

BW
DFB1 Open source 50MHz 030 and TT-RAM accelerator for the Falcon
DSTB1 Open source 16Mhz 68k and AltRAM accelerator for the ST
Smalliermouse ST-optimised USB mouse adapter based on SmallyMouse2
FrontBench The Frontier: Elite 2 intro as a benchmark
User avatar
frank.lukas
Posts: 660
Joined: Fri Jan 19, 2018 11:52 am

Re: DFB1r5 Design Review Thread

Post by frank.lukas »

Are there any ideas for handling the FPU into the Firmware? In the PAK GALs, the FPU_CS is simple. Doesn't that work not in the DFB1 firmware?

fpucs = fc1 * fc0 * !a19 * !a18 * a17 * !a16 * jp3; ´FPU enabled
avec = fc1 * fc0 * vpa;
User avatar
Badwolf
Posts: 2231
Joined: Tue Nov 19, 2019 12:09 pm

Re: DFB1r5 Design Review Thread

Post by Badwolf »

frank.lukas wrote: Mon Jun 27, 2022 6:05 pm Are there any ideas for handling the FPU into the Firmware? In the PAK GALs, the FPU_CS is simple. Doesn't that work not in the DFB1 firmware?

fpucs = fc1 * fc0 * !a19 * !a18 * a17 * !a16 * jp3; ´FPU enabled
avec = fc1 * fc0 * vpa;
Hi Frank,

Sorry -- what are you asking here? FPU is already in the firmware & the question doesn't seem to follow on from the (old) thread above.

BW
DFB1 Open source 50MHz 030 and TT-RAM accelerator for the Falcon
DSTB1 Open source 16Mhz 68k and AltRAM accelerator for the ST
Smalliermouse ST-optimised USB mouse adapter based on SmallyMouse2
FrontBench The Frontier: Elite 2 intro as a benchmark
Post Reply

Return to “HARDWARE DISCUSSIONS”